High load using mysqlump utilty

Hello,

I have a big vBulletin forum, with about 4-5k users online average, reaching max peaks of 8k online - 3.000.000 messages, about 100k registered users, and growing... This is the server structure:

1 Webserver (Xeon 3.2 x4 procs, 4 GB RAM, 2 x 250 GB SATA, plain RHEL 5.4)
1 MySQL Server (Xeon 3.2 x4 procs 4GB RAM, 2 x 250 GB SATA, plain RHEL 5.4)
1 Images/Template Server (Xeon 3.2 x4 procs 4GB RAM, 2 x 250 GB SATA, plain RHEL 5.4)

MySQL handles 2 dbs:

Code: 
4.6G    /var/lib/mysql/site_com
1.7G    /var/lib/mysql/site_net
However, we have a problem when we run daily night backups, this is the way I do the backups on the script:

Code: 
for i in $(mysql -u root -p$MYSQLPASS -Bse 'show databases' | egrep -v 'test|information'); 
do mysqldump --opt -p$MYSQLPASS $i -c> $TMPDIRSQL/mysql-dump-$i-$DATE.sql; done
It takes about 20 minutes to do the backup.. while it's still serving request from the WebServer, anyway, from minute 5 until it finishes, the load average is getting higher and higer until it goes down.

This is my my.cnf in case it helps:

[mysqld]
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
skip-locking
skip-innodb
skip-bdb
query_cache_limit=1M
query_cache_size=48M
query_cache_type=1
max_connections=1200
interactive_timeout=100
wait_timeout=10
connect_timeout=10
thread_cache_size=128
key_buffer=48M
join_buffer=8M
max_allowed_packet=16M
table_cache=2036
sort_buffer_size=1M
read_buffer_size=1M
read_rnd_buffer_size=2M
max_connect_errors=10

# Try number of CPU's*2 for thread_concurrency
thread_concurrency=4
myisam_sort_buffer_size=64M

# Add
max_heap_table_size = 48M
tmp_table_size = 48M
low_priority_updates=1
concurrent_insert=2

[mysql.server]
user=mysql
#basedir=/var/lib

[mysqld_safe]
err-log=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid
log-slow-queries=/var/log/slow-queries.log
Any ideas are appreciated!

Thanks!

 

 

 

 

Top